CARBON DIOXIDE SEQUESTRATION UNDERGROUND LASER BASED
DETECTION SYSTEM
by
Jamie Lynn Barr
A thesis submitted in partial fulfillment
of the requirements for the degree
of
Master of Science
in
Electrical Engineering
MONTANA STATE UNIVERSITY
Bozeman, Montana
November, 2009
c Copyright
by
Jamie Lynn Barr
2010
All Rights Reserved
ii
APPROVAL
of a thesis submitted by
Jamie Lynn Barr
This thesis has been read by each member of the thesis committee and has been
found to be satisfactory regarding content, English usage, format, citations, bibliographic style, and consistency, and is ready for submission to the Division of Graduate
Education.
Dr. Kevin S. Repasky
Approved for the Department of Electrical Engineering
Dr. Robert Maher
Approved for the Division of Graduate Education
Dr. Carl Fox
iii
STATEMENT OF PERMISSION TO USE
In presenting this thesis in partial fulfullment of the requirements for a master’s
degree at Montana State University, I agree that the Library shall make it available
to borrowers under rules of the Library.
If I have indicated my intention to copyright this thesis by including a copyright
notice page, copying is allowable only for scholarly purposes, consistent with “fair
use” as prescribed in the U.S. Copyright Law. Requests for permission for extend
quotation from or reproduction of this thesis in whole or in parts may be granted
only by the copyright holder.
Jamie Lynn Barr
November, 2009
iv
ACKNOWLEDGEMENTS
I would like to thank my advisor Kevin Repasky for giving me this opportunity and for all of his guidance throughout the process. I would also like to
thank John Carlsten for his many helpful insights and always being available
for help. Thanks to Seth Humphries and Amin Nehrir for their guidance and
assistance. Finally I would like to thank my parents for always supporting me
in all that I try to do.
This work was kindly supported by the Department of Energy under Award
No. DE-FC26-04NT42262. However, any opinions, findings, conclusions, or recommendations expressed herein are those of the author(s) and do not necessarily
reflect the views of DOE.
v
TABLE OF CONTENTS
1
INTRODUCTION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
1
Carbon Sequestration . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
ZERT Field Site . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
1
5
THEORY . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
7
Mathematical Calculation for finding CO2 Concentration . . . . . . . . .
9
SUMMER 2008 DIFFERENTIAL ABSORPTION INSTRUMENT . . . .
14
System Setup . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Photonic Bandgap Fiber . . . . . . . . . . . . . . . . . . . . . . . . . . .
Experimental Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
14
19
21
SUMMER 2009 DIFFERENTIAL ABSORPTION INSTRUMENT . . . .
26
Instrument Improvements . . . . . . . . . . . . . . . . . . . . . . . . . . .
First 2009 Field Experiment . . . . . . . . . . . . . . . . . . . . . . . . .
Second 2009 Field Experiment . . . . . . . . . . . . . . . . . . . . . . . .
26
29
31
CONCLUSION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
34
Future Work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
34
REFERENCES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
36
2
3
4
5
vi
LIST OF TABLES
Table
Page
1
Worldwide capacity of potential CO2 storage reservoirs . . . . . . . .
2
2
Molecular data from HITRAN for absorption features presented in
Figure 4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
9
3
Specifications for Hollow Core Photonic Crystal Fiber . . . . . . . . .
20
4
Splicing parameters for fusion splicing a PBG to SMF . . . . . . . . .
21
vii
LIST OF FIGURES
Figure
1
2
3
4
5
6
7
Page
Monthly average CO2 concentrations in ppm, taken continuously since
1958 to today from several observatories around the world. The black
dots represent measured data with the black lines connecting them
with a curve fit to the data. The monitoring sites are located at the
South Pole (SPO), Samoa (SAM),Christmas Island (CHR), Mauna
Loa, Hawaii (MLO), La Jolla, California (LJO), andPoint Barrow,
Alaska (PTB)[1] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
3
Areal view of the ZERT field experiment with various experiments as
well as the location of the 100 meter injection well. . . . . . . . . . .
6
Transmission of light as a function of wavelength around 2 microns.
The absorption features are associated with all molecules in the atmosphere [2] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
8
A more isolated section of absorption features of CO2 and water vapor
around 2 µm [2] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
8
Transmission of light as a function of wavelength. This is the zoomed in
version of Figure 4 with the desired absorption features to be scanned
over. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
10
Predicted percent transmission as a function of concentration for an
independent CO2 absorption feature . . . . . . . . . . . . . . . . . . .
12
Prediction of the error associated with the calculation of CO2 concentration values . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
12
8
Power output of the 2µm DFB laser with a threshold current of 19.5 mA 15
9
Bread board setup for launching 2 µm DFB diode laser light into a
single mode fiber splitter . . . . . . . . . . . . . . . . . . . . . . . . .
16
Schematic of the differential absorption instrument for measuring underground CO2 concentrations for the 2008 field release . . . . . . . .
17
A picture of the underground portion of the sensor. Gas permeable
membranes allow the CO2 to enter this sensor while keeping out the
water and dirt. This box corresponds to box 2 shown schematically in
Figure 10 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
18
Plot of the normalized transmission as a function of wavelength for the
three underground sensors . . . . . . . . . . . . . . . . . . . . . . . .
19
10
11
12
viii
LIST OF FIGURES – CONTINUED
Figure
Page
13
Dry ice test to determine the diffusion time for the PBG fiber. . . . .
22
14
Underground CO2 concentration as a function of time. The solid black
(dashed red) line indicates measurements made over the (1 m lateral
from) the injection pipe. Rain events that occured during this release
affected the diffusivity of the soil causing the large fluctuations in the
underground CO2 concentrations seen in the plot. . . . . . . . . . . .
23
A plot of the CO2 concentration as a function of time. The vertical
lines indicated rain events that can affect the soil diffusivity causing
variations in the underground transport of the CO2 . [3] . . . . . . . .
24
15
16
A comparison of the CO2 concentration measured using the open path
absorption cell and the PBG fiber. Good agreement between these two
measurements indicates that the PBG fiber is a valid sensor technology 25
17
Power output for the fiber coupled 2µm DFB diode laser . . . . . . .
27
18
Schematic for the fiber optical switch. It allows for optical power to
be delivered to different output channels . . . . . . . . . . . . . . . .
28
19
Schematic for the 2009 field release . . . . . . . . . . . . . . . . . . .
29
20
Results for an in-lab experimental test run using dry ice. . . . . . . .
30
21
CO2 concentrations for the month long injection experiment. The solid
black line indicates measurements made over the injection well, while
the solid red line indicates measurements made 2 meters away from
the pipe. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
31
CO2 concentrations for the week long injection experiment. The solid
black line indicates measurements made by the 1 meter free space cell
over the injection well while the solid red line indicates measurements
made by the 1 meter fiber cell, and the green line indicates the fiber
cell 1 meter away from the pipe. . . . . . . . . . . . . . . . . . . . . .
33
22
ix
ABSTRACT
Carbon dioxide (CO2 ) is a known greenhouse gas. Due to the burning of fossil
fuels by industrial and power plants the atmospheric concentration of CO2 has been
rising over the past 50 years. Carbon capture and sequestration provides a method to
prevent CO2 from being emitted into the atmosphere. Successful carbon sequestration
will require the development of many pieces of technology including development of
monitoring tools and techniques. An underground laser based monitoring system was
built and tested at Montana State University (MSU) to measure sub-surface CO2
concentrations at a sequestration site. The instrument uses differential absorption
spectroscopy by temperature tuning a distributed feedback diode laser over several
CO2 absorption features located at 2.004 microns. The instrument utilizes photonic
bandgap fibers for sub-surface spectroscopy CO2 concentration measurements. The
instrument was tested at a controlled release facility located on the MSU campus. The
field and CO2 release are managed by the Zero Emissions Research and Technology
group at MSU. Three CO2 injection tests were done over the coarse of two summers
to simulate a fault or fracture line at a sequestration site. Results from all three tests
are presented showing that the underground differential absorption instrument could
be used to monitor sequestration sites.
1
INTRODUCTION
It has been documented that atmospheric concentrations of CO2 have been changing since the 1950’s when Charles David Keeling began measuring CO2 levels in the
atmosphere [1]. Atmospheric CO2 concentrations have increased from 310 ppm in
1957 to 387 ppm in 2008 [1]. Figure 1 shows a plot of the measured CO2 concentrations at multiple locations around the world [1]. The figure shows the seasonal cycle
that occurs naturally, this cycle occurs from plant and microbes that take up and
give off CO2 at varying rates on daily and seasonal time periods [4]. Along with the
seasonal fluctuation of CO2 there is an exponential increase in CO2 concentrations
that is visible since the study began.
The increasing levels of atmospheric CO2 can potentially impact the global climate
through an enhancement of the greenhouse effect[5],[6],[7],[8],[9]. The greenhouse
effect results from the fact that CO2 allows the short wave solar radiation to pass but
absorbs long wave thermal radiation, trapping heat in the atmosphere [9].
The leading cause of the increased atmospheric CO2 comes from the burning of
fossil fuels including coal, oil, and natural gas as well as deforestation [5],[6]. These
atmospheric concentrations are raising concerns about the impacts it has on the global
climate [5],[6],[7],[8],[9]. In order to counteract the effect that human contributions
have made to that increasing CO2 concentrations a plan to capture and store CO2
has been proposed.
Carbon Sequestration
Most climate forcing greenhouse gas emission facilities have been produced within
the last 50 years with 70% of the anthropogenic increase occurring after 1950 [9]. If
2
Table 1: Worldwide capacity of potential CO2 storage reservoirs
Sequestration option
Worldwide Capacity (order of magnitudes)
(gigatons of carbon)
Ocean
103 GtC
2
Deep Saline Formations
10 − 103 GtC
Depleted Oil and Gas Resevoirs
102 GtC
Coal Seams
10-102 GtC
Terrestrial
10 GtC
CO2 emissions are not regulated by the year 2100 the measured atmospheric CO2
level will increase to 737 ppm which is more than double the concentration level in
1990 [9]. Carbon Sequestration is one method to reduce the emission of CO2 into
the atmosphere. Carbon sequestration directly captures the CO2 produced by industrial and utility plants and stores it in secure underground reservoirs [10],[11],[12].
Geological sinks for CO2 include deep underground saline formations, depleted oil
and natural gas reservoirs, and coal seams that are unable to be mined [12]. Taking
all of these areas into consideration the underground geological formations can hold
hundreds to thousands of gigatons of CO2 . Table 1 is an estimate of the worldwide
capacity of potential CO2 storage [12].
The technology that would be used to transfer CO2 to the underground sites is
well developed. Before CO2 gas can be sequestered from power plants it must be
captured. CO2 is captured and separated as a by-product from industrial processes
such as synthetic ammonia production; it is then compressed into liquid form so that
it can be transported underground [8]. An example where this technique is currently
being used is in enhanced oil recovery; CO2 is injected underground into geological
formations that are known to house oil [12]. In 1998, 43 million metric tons of CO2
were injected at 67 commercial enhanced oil recovery projects in the United States
3
Figure 1: Monthly average CO2 concentrations in ppm, taken continuously since
1958 to today from several observatories around the world. The black dots represent
measured data with the black lines connecting them with a curve fit to the data.
The monitoring sites are located at the South Pole (SPO), Samoa (SAM),Christmas
Island (CHR), Mauna Loa, Hawaii (MLO), La Jolla, California (LJO), and Point
Barrow, Alaska (PTB)[1]
.
4
[12]. Although CO2 is being used in enhanced oil recovery more CO2 is produced by
factories than can be used for this purpose.
Occurring around the world are four large scale carbon sequestration projects
that are used for learning and demonstration purposes to house CO2 until alternate
solutions to the production of greenhouse gas emission can be made. These sites
include the North Sea in Norway, the Weyburn project in Canada, the Frio experiment
in Texas, and the In Salah project in central Algeria [11], [12], [13], [14], [15], [16].
The Sleipner and SACS projects in the North Sea has been in operation since 1996
and by 2004 8 million tons of CO2 have been injected into an underwater reservoir
[15]. The Weyburn project is a large scale demonstration of carbon sequestration
in southeastern Saskatchewan. For this demonstration 5000 tons of CO2 are being
injected into the underground reservoir daily with the ultimate goal of sequestering
20 million tons of CO2 [11]. The Frio experiment utilizes large saline formations
underneath the United States Gulf Coast. For this experiment 1600 metric tons of
CO2 were injected into the underground reservoir and the migration and leakage of
the site was closely monitored [14]. The In Salah project injects CO2 into the Krechba
Carboniferous Sandstone reservoir via three long horizontal wells at a depth of 1900
m. To date the In Salah project has injected 2.5 million tons of CO2 underground
with an estimated capacity of 14 million tons over the lifetime of the sequestration
project [16].
For carbon sequestration to be successful in terms of eliminating the increasing
CO2 concentrations in the atmosphere, seepage rates of less than 0.1% of injected
volume per year need to be maintained [17]. The main areas to watch for leakage
in large scale projects are leaking injection wells, leakage from improperly sealed
abandoned wells, and leakage through geologic faults and fractures. To insure that
the CO2 is captured successfully, monitoring of the sites is needed.
5
ZERT Field Site
A small scale simulation test site was created to allow further exploration in the
monitoring of underground sequestration sites. The Zero Emission Research Technology (ZERT) field site at Montana State University (MSU) was developed for testing
surface monitoring devices [4]. The Zert field site is a 30 acre agricultural field on the
campus of MSU. A 100 meter long horizontal well was installed about 2 meters below
ground. This horizontal pipe was created with a 70 meter long center section that is
designed with gaps in the pipe so that the site can be used to simulate a linear fault
line or fracture. The 70 meter long section is then divided into 6 separate sections
using an inflatable packer system that controls the rate of flow for each individual
section [4]. The ZERT field Site can be seen in Figure 2 [18]. The underground horizontal injection well is represented by the solid black line. The make up of the field
consists of a thick sandy gravel and cobble deposits with a mixture of silts and clay
as the topsoil [4]. The field is also covered in prairie grasses, alfalfa, and Canadian
thistle. An important feature of the field site is that the water table depth for the
site sits at about 1.6 meters below ground meaning that the pipe that releases the
CO2 is located below the water table.
The goal of the ZERT field site is to investigate various monitoring systems to
determine the most productive way in which to monitor sequestration sites. Multiple
research labs from around the country are present each year at the site in order to
test their systems [19]. Many different ideas are tested including, but not limited
to: plant life monitoring, CO2 flux measurements, the use of eddy covariance towers,
ground water monitoring of ph levels, as well as differential absorption instruments
[19]. Figure 2 shows the placement of some of these experimental projects. The below
ground differential absorption instrument will be highlighted in this paper.
6
Figure 2: Aerial view of the ZERT field experiment with various experiments as well
as the location of the 100 meter injection well.
7
THEORY
Optical methods can be used to monitor CO2 levels in the atmosphere, by tuning
over absorption features associated with the molecule of interest [19]. By choosing
the right absorption feature, the CO2 concentration can be calculated. Presented in
this section is the theory behind how the concentration of CO2 can be calculated.
The model HITRAN was used to compare measured results and data obtained by
the underground differential absorption monitoring instrument. HITRAN is a highresolution transmission molecular absorption database that was created by the Air
Force Cambridge Research Laboratories (AFCRL) [2]. It accurately depicts how light
is absorbed by molecules in the atmosphere. From this model a group of absorption
features were chosen. Figure 3 shows the absorption features depicted by HITRAN
around 2 µm. Two main criteria were used to determine what wavelength the laser
should be tuned over. The first criterion was that a strong absorption feature was
needed for a short path length that was isolated from other molecular absorption
features. The second criterion was the availability of a distributed feedback (DFB)
laser that would tune over the desired wavelengths.
HITRAN was used to see where absorption features are available without any
interference from other molecules. Figure 3 shows a range of wavelengths around 2µm
that has many CO2 absorption features without interference from other molecules.
Absorption lines around 2µm were chosen because many independent CO2 absorption
lines can be found with minimal interference of other molecules while still having the
availability of a tunable laser [19]. Figure 4 shows a narrowerer spectral band where
absorption features that include water vapor and CO2 are present. To accompany
8
Figure 3: Transmission of light as a function of wavelength around 2 microns. The
absorption features are associated with all molecules in the atmosphere [2]
Figure 4: A more isolated section of absorption features of CO2 and water vapor
around 2 µm [2]
9
Table 2: Molecular data from HITRAN for absorption features presented in Figure 4
Molecule Wavelength λ(µm) Line intensity S(cm−1 mol−1 )
CO2
2.004548
1.322E-21
CO2
2.004545
1.463E-23
H2 O
2.004544
2.319E-23
H2 O
2.004532
6.080E-24
H2 O
2.004493
1.909E-23
CO2
2.004491
1.618E-23
CO2
2.004137
1.171E-23
CO2
2.004048
1.306E-23
CO2
2.004019
1.332E-23
CO2
2.003615
1.036E-23
CO2
2.003503
1.302E-21
CO2
2.002998
1.241E-21
H2 O
2.002828
4.490E-23
Figure 4 a table of values for the absorption lines that will be used to detect CO2
with the underground monitoring instrument is shown [19].
Table 2 includes the wavelength of the absorption feature, the line intensity, S,
of the individual line, as well as the molecule that is present. Narrowing the search
even more, a group of four absorption features was chosen. Figure 5 shows the four
absorption lines that were chosen for use in this experiment [19]. These four lines
do not have any other molecular absorption interference and by carefully researching
the availability of diode lasers in the infrared, a Nanoplus 2 µm distributed feedback
diode laser was able to tune over all four lines. A path length of 1 meter was used
for the calculation.
Mathematical Calculation for finding CO2 Concentration
The optical method uses the fact that CO2 absorbs light at specific wavelengths
and the amount of absorbed light can be used to determine the concentration level
10
Figure 5: Transmission of light as a function of wavelength. This is the zoomed in
version of Figure 4 with the desired absorption features to be scanned over.
of CO2 . The amount of light transmitted through a path length L can be found from
the relationship:
T =
I
= e−αL
I0
(1)
where T is the transmission, I is the optical intensity after the path length L, and I0
is the intensity at the beginning of the path length. Equation 1 can be rewritten to
isolate the absorption per unit length with respect to transmission:
−αL = ln T
(2)
The absorption per unit length is α [cm−1 ] and when multiplied by the path length
it can also be written in terms of the molecular line intensity S [cm/molecule] and
normalized line shape g(υ - υ 0 ) [cm]. This can be rewritten as:
11
αL = Sg(υ − υo )N Pa L
(3)
where N [molecule/(cm3 atm)] is the number density of the molecules of interest and
Pa [atm] is the partial pressure of the molecules of interest. The factor N, the number
density, is scaled by temperature and can be written as:
N = NL
296
Ta
(4)
where NL =2.478x1019 [molecules/(cm3 atm)] is Loschmidt’s number and Ta is the atmospheric temperature in Kelvin. The total number density of molecules of interest
can be written as Ntotal =NL PT [molecules/cm3 ] where PT is the total pressure in
[atm]. Therefore by reconfiguring equations (2-4) an equation for the concentration
C [ppm] of the molecule of interest can be written as:
C = 106
Pa
− ln T
= 106
PT
Sg(υ − υ0 )NL ( 296
)PT L
Ta
(5)
The calculated concentration is dependent on the ambient temperature, barometric
pressure and the path length of the light through the atmosphere. Equation (5) was
used to calculate all of the measured results that will be presented in this paper [19].
A plot of the transmission as a function of CO2 concentrations is shown in Figure
6. For this calculation the following parameters were used, a path length of 1 meter,
a total pressure of 1 atm, and an atmospheric temperature of 287.2 K. As the concentration of CO2 increases in the atmosphere the amount of light able to be transmitted
decreases in an exponential fashion. Along with understanding how the concentration
level relates to the transmission of light, a plot of CO2 concentration with respect to
percent error was created to understand the error associated with our measurements.
12
Figure 6: Predicted percent transmission as a function of concentration for an independent CO2 absorption feature
Figure 7: Prediction of the error associated with the calculation of CO2 concentration
values
13
By considering a 2% inaccuracy with the measurement of the transmission of light,
the understanding of how accurate the calculated value of concentration of CO2 is
shown in figure 7. When the concentration of CO2 is low small changes greatly effect
the level of transmission but as the absorption features become more prominent more
accurate calculations can be made. It should be noted that the measurements made
out in the ZERT field site ranged from 6000 ppm to no higher than 160,000 ppm.
This relates to a very small scale error prediction.
14
SUMMER 2008 DIFFERENTIAL ABSORPTION INSTRUMENT
Using the molecular absorption described in chapter 2, an instrument was built to
measure CO2 concentrations. A distributed feedback (DFB) diode laser was purchased from Nanoplus to be used to scan over multiple absorption features present
around the 2.004 µm wavelength. A DFB laser was chosen because of its tuning
characteristics [20]. A DFB laser uses a grating as part of the gain material. A stable
center wavelength is set by manufacturing the spacing of the grating, and tuning is
achieved through thermal expansion and contraction of this grating allowing the DFB
diode laser to be tuned over several CO2 absorption features. An internal thermoelectric cooler (TEC) was integrated in the DFB laser package in order to tune the
laser over the desired absorption features. The optical output of the laser was tested
to determine the optimal operation level at which to run the laser during the experiment. The temperature of the laser was set to a constant temperature of 23 C. The
current to the diode was changed while the output of the laser was measured. The
measured power as a function of laser drive current can be seen in Figure 8. The
lasing threshold of this laser occurs when the output voltage of the diode significantly
increase which is at about 19.5 mA for this particular laser.
System Setup
Once the laser specifications were understood a breadboard configuration was
constructed in order to launch light into a single mode fiber (SMF-28E) with FCAPC connectors. Two mirrors were used to direct the light coming out of the DFB
diode laser into the fiber. The two mirrors allow vertical and horizontal steering
capabilities needed to achieve good optical coupling into the optical fiber. The fiber
15
Figure 8: Power output of the 2µm DFB laser with a threshold current of 19.5 mA
was placed into a tip-tilt stage that housed a microscope objective lens to focus the
light down so that the optimal amount of was incident on the end of the fiber. The
light was then sent through a single mode in-line fiber optic splitter that allowed the
light from the DFB laser to be sent to two different channels. Figure 9 shows the
bread board setup for this system.
A schematic of the below ground instrument is shown in Figure 10. The light
transferred to the in-line optical fiber splitter is split in such a way that approximately
half the light is transferred to box 1 and the rest of the light is transferred to box 2.
Inside the boxes are specific length cells that allow the laser light to travel along a
desired path onto a detector. The two underground water proof boxes were configured
to allow soil gases to penetrate through the box but not allow water to seep into the
boxes that were buried underground. This was accomplished by placing millipore
filters on the outside of the below ground boxes. Millipore filters are manufactured
to only allow certain sized molecules through, therefore gas molecules can penetrate
through the filter but it does not allow water molecules to seep in [21]. Once the filters
were placed onto the boxes the boxes were fitted with BNC connectors so that power
can be delivered to the box and signal from the detector can be collected. A FC-APC
16
Figure 9: Bread board setup for launching 2 µm DFB diode laser light into a single
mode fiber splitter
17
Figure 10: Schematic of the differential absorption instrument for measuring underground CO2 concentrations for the 2008 field release
connecter was also attached to the underground box in order to transfer light into
the boxes. Once all the external connections and filters were applied the boxes were
sealed with a waterproof silicone sealant. Figure 11 shows the configuration of box 2.
Once the boxes were constructed the rest of the system was setup in the following
manner. Light from one of the channels of the first in-line fiber optic splitter is
transferred to box 1. Light transferred to box 1 is then split again using another
in-line optical splitter. Half of the light is transferred to a reference detector D1 while
the other half of the light is transferred to a one meter long free space cell; light from
the single mode fiber is incident onto a lens that focuses the light one meter away.
The detector, D2, is placed one meter away on an adjustable mount. Light from the
second channel of the first in-line fiber optic splitter is transferred to box 2. Light in
box 2 is again coupled into an in-line fiber optic splitter. Located in box 2 is another
free space cell that was created in a similar fashion as the free space cell located in
box 1, but instead of having a path length of one meter the cell was constructed to be
0.3 meters in length. This light is then incident on the detector, D3. The other half of
18
Figure 11: A picture of the underground portion of the sensor. Gas permeable membranes allow the CO2 to enter this sensor while keeping out the water and dirt. This
box corresponds to box 2 shown schematically in Figure 10
the light from the in-line optical splitter in box 2 is transmitted through single mode
fiber that has been spliced to a 1 meter length of photonic bandgap fiber (PBG). The
light exiting the PBG fiber is then incident onto the final detector D4.
The data collected by the underground sensors is done in the following way. The
operating temperature, which controls the wavelength of the DFB diode laser, is set
by a computer controlled temperature controller. The four detectors voltages are
then recorded by the computer using a multichannel voltmeter. The computer then
steps the operating temperature of the laser, which changes the wavelength, and the
detector voltages are again recorded by the computer. This process is repeated generating a scan over the CO2 absorption features that were presented in the previous
section. The normalized transmission spectra is then calculated by dividing the voltages read by detectors D2, D3, and D4 by the reference detector D1. The normalized
transmission spectra are then used to calculate the concentration level of CO2 using
the equations presented earlier. A plot of the three normalized transmission spectra
19
Figure 12: Plot of the normalized transmission as a function of wavelength for the
three underground sensors
can be seen in figure 12. The plot shows that all three sensor scan over the four
absorption features that are desired, the measurements made for each sensor were
done for different concentration levels. Once the concentration values are calculated
the computer records the data with respect to time.
Photonic Bandgap Fiber
The use of the PBG fiber in place of the rigid free space cells allows the system
to have a meter long absorption cell that can be coiled into a much more compact
underground container. The placement of the PBG fiber in the same box as a 0.3
meter absorption cell was to validate the effectiveness of the PBG fiber for future
use. PBG fibers are low loss waveguides that use a honeycomb configuration to guide
light down a center air capillary [22]. The specifications for the hollow core photonic
bandgap fiber used in this instrument can be seen in table 3. To transmit light into
the PBG fiber a program was made to splice the hollow core PBG fiber to a traditional
20
Table 3: Specifications for Hollow Core Photonic Crystal Fiber
Operation Specifications
Item
Specification
Unit
Center Operating wavelength
2.0
µm
Transmission Spectrum
1.9-2.1
µm
Material
Fused Silica
Attenuation at Operating Wavelength
less than 0.3
db per m
Percent of light propagating in air (1)
greater than 95%
Effective mode index
0.995
Key Geometric Specifications
Item
Specification
Unit
Nominal Core Diameter
12
µm
Nominal Fiber Diameter
140
µm
Nominal Lattice Spacing (Pitch)
3.6
µm
Air Filling Fraction in the holey region
greater than 0.90
Nominal Coating Diameter (single layer acrylate)
approx. 250
µm
glass core single mode fiber [23]. The PBG fiber was spliced to SMF-28e single mode
fiber using an Ericsson FSU-995 electric fusion splicer. Considerations for creating a
splicing program for the hollow core fiber to single mode fiber were that the strongest
splice possible was wanted without collapsing the hollow center core. To achieve this,
the parameters used for the fusion splice on the Ericsson FSU-995 were set to those
shown in table 4. The program was successfully used to fuse the two fibers together.
With multiple splices performed a maximum through put of about 10% was achieved.
This was enough light output to successfully scan over the desired absorption features
needed.
Next a test was conducted using dry ice in order to determine the validity of
PBG fiber in detecting the presence of CO2 . The 0.3 meter cell along with the PBG
fiber cell were placed into box 2 along with a small amount of dry ice. The test
was performed over a 40 hour period in order for the dry ice to evaporate and a
rise and fall of CO2 levels can be documented. Figure 13 shows the concentration
21
Table 4: Splicing parameters for fusion splicing a PBG to SMF
Splicing Parameters
Prefuse time
0.2 seconds
Prefuse Current
10.0 mA
Gap
15.0 µm
Overlap
12.0 µm
Fusion time 1
0.3 seconds
Fusion Current1
10.5 mA
Fusion time 2
9.0 seconds
Fusion Current 2
10.0 mA
Fusion time 3
3.0 seconds
Fusion Current 3
9.5 mA
Offset
+260
levels that the 0.3 meter cell was able to measure along with a plot of the PBG fiber
measurements. It takes longer for the PBG fiber to reach the same levels as the 0.3
meter cell because there is a period of time that is needed for the CO2 to diffuse
into the end of the PBG fiber. With this experiment it was calculated that it takes
about 4 hours for the CO2 to diffuse into the PBG fiber at the 1/e value. This relates
favorably to the results found in Hoo et al. that estimates a 200 min diffusion time
for C2 H2 gas into a 1 meter length of PBG fiber [24]. With this knowledge the use of
the PBG fibers could be used with the knowledge that there is a lag in detection of
CO2 due to the diffusion into the fiber.
Experimental Results
A one month experimental release was done at the ZERT field site at MSU from
July 9th to August 7th, 2008. The two boxes were positioned in the following way
at the field site. Box 1, which houses the reference detector and the 1 meter long
free space cell, was located 1 meter away from the pipe. The second box was located
directly over the injection pipe. Each of the boxes was buried approximately 0.7
22
Figure 13: Dry ice test to determine the diffusion time for the PBG fiber.
meters below ground just above the cobble layer. The first plot shown in Figure
15 represents the two free space cells that were buried. Figure 14 shows the CO2
concentration for these two cells as a function of time. The solid line represents
measurements made using the 0.3 m absorption cell that was located over the release
pipe while the dashed line represents the 1 m absorption cell that was located 1 m away
from the pipe. The system was set in place on July 4th and was left underground until
August 13th. The vertical green line represents the beginning of the CO2 injection
and the vertical black line represents the end of the injection of CO2 . The system
was placed underground before the injection started to measure the background CO2
concentrations that are produced by plant and microbial respiration. About 24 hours
after the start of the underground injection the 0.3 m cell begins to see an increase in
underground CO2 concentration. About 64 hours after the start of the underground
injection the 1 m cell begins to see an increase in the underground CO2 concentrations.
23
Figure 14: Underground CO2 concentration as a function of time. The solid black
(dashed red) line indicates measurements made over the (1 m lateral from) the injection pipe. Rain events that occured during this release affected the diffusivity of the
soil causing the large fluctuations in the underground CO2 concentrations seen in the
plot.
Figure 14 shows an interesting feature in that it takes about 40 hours for the CO2
that is released to move 1 m laterally from the pipe.
During the month long experiment multiple rain events occurred. A close up
look at a small section of time in which significant rain events occurred can be seen
in Figure 15. Figure 15 is a plot of the CO2 concentration as a function of time.
Multiple rain events are represented in this plot as vertical bars. The rain events
caused the soil to become wet affecting the diffusivity of the CO2 through the soil.
Once rain events occur, there is a period of time when the CO2 concentration would
significantly drop, almost down to background levels.
As stated before the PBG fiber was placed into the same underground box as
the 0.3 meter absorption cell, this was done to validate the use of these fibers as
successful absorption cells in the future. By placing the two cells in the same box a
24
Figure 15: A plot of the CO2 concentration as a function of time. The vertical
lines indicated rain events that can affect the soil diffusivity causing variations in the
underground transport of the CO2 . [3]
comparison to the concentration levels could be made. This comparison is shown in
figure 16. The black line represents CO2 concentration measurements made with the
0.3 m open path cell while the red line represents CO2 concentration measurements
made with the PBG fiber sensors. Figure 16 shows a strong correlation between the
two cells. The close up portion of this plot also shows that the system is sensitive
enough to measure the daily diurnal cycle that occurs naturally from the plants and
microbes located in the soil. The diurnal cycle for the PBG fiber lags that of the 0.3
meter cell which is what we expected after we ran initial tests. This lag time results
from the slow diffusion rate of the CO2 into the hollow core of the PBG fiber. The
strong correlation between the two cells confirms that the use of PBG fiber in the
future would be a beneficial way of shrinking the size of the underground portion of
the system while still keeping the integrity of the 1 meter path length.
25
Figure 16: A comparison of the CO2 concentration measured using the open path
absorption cell and the PBG fiber. Good agreement between these two measurements
indicates that the PBG fiber is a valid sensor technology
26
SUMMER 2009 DIFFERENTIAL ABSORPTION INSTRUMENT
The initial deployment for the underground system using the PBG fiber was
discussed in Chapter 3. The initial findings concluded that the system worked successfully and after minor adjustments the system could be expanded to incorporate
multiple point source fiber sensors.
Instrument Improvements
One of the first improvements made to the new system was to incorporate a
new laser into the system. A fiber coupled 2 µm DFB diode laser with an internal
TEC is used. The 2 µm DFB fiber coupled laser was purchased from Nanoplus and
allows for light exiting the laser to be already coupled into an optical fiber with an
FC/APC adaptor [20]. Once the new laser was received the power output was tested
to understand the best operating parameter to use. Figure 17 shows the power output
of the laser with respect to the current that drives the system. By incorporating the
new laser there was no longer a need for a bread board configuration to fiber couple
light. The bread board configuration that was used in the 2008 field experiment was
subject to optical power fluctuations due to thermal expansion and contraction of
the mounted mirrors that are used to launch light into the fiber. In eliminating the
mirrors a more temperature stable system was created.
The next addition to the new system was a MEMS fiber optic switch. The switch
was purchased from Pickering interfaces and was used to optimize the amount of power
delivered to each underground absorption cell. The optical switch uses a MEMS based
mirror system to switch optical light form one output channel to the next. This means
that all of the light is directed to one underground cell at a time. This system allows
27
Figure 17: Power output for the fiber coupled 2µm DFB diode laser
for the maximum amount of power to be delivered to the underground absorption
cells. Figure 18 shows the general concept behind the 4 channel fiber optical switch.
The focus of the 2009 field experiment was to deploy an array of PBG fibers to map
out the movement of CO2 leaking from the underground injection well. Three new
PBG fiber absorption cells were produced to be placed in new underground boxes.
The new underground boxes were 1/3 of the size of the previous years experiment
with only the PBG fiber sensor to be deployed in each of the boxes. Each box was
again outfitted with Millipore filters to allow molecular gasses to penetrate through
the boxes. The BNC connectors and fiber adaptor were configured in such a way that
would allow for PVC piping to fit around them and protect the fiber and coax cables
from bending and kinking.
A schematic for the below ground instrument for the 2009 field experiment is
shown in Figure 19. Light for the fiber coupled 2 µm DFB laser is connected to an
in-line fiber splitter that sends 10% of the light to a reference detector D1 and 90% of
the light to the MEMS fiber optical switch. Light is then transferred to box 1 where
a 1 meter long free space absorption cell is located. The signal is then recorded using
28
Figure 18: Schematic for the fiber optical switch. It allows for optical power to be
delivered to different output channels
the Multi-Channel Voltmeter. Once the scan is completed over the four absorption
features the concentration level of box 1 is calculated. Next the MEMS optical switch,
which is controlled by the computer, is switched to the next channel. Channel two
sends light to box 2 which houses a 1 meter long fiber optical absorption cell. The 1
meter fiber cell is incident on detector D3 and the data is recorded by the voltmeter.
Again the concentration level is calculated and the optical switch is changed to the
next channel. The process is repeated for the next two channels. Each scan takes
approximately 11.5 minutes, therefore the switch returns to channel 1 (box 1) every
46 minutes.
To test the validity of the system an experiment was done using dry ice. The
dry ice was placed in all four underground sensor boxes which simulates the expected
elevated CO2 concentrations. The system was run for approximately 48 hours so that
the rise and fall of CO2 concentrations could be mapped for all four sensors. Figure
20 shows how each sensor could see the increase in CO2 concentration and how it
each sensor was sensitive enough to see a gradual decrease in concentration due to
29
Figure 19: Schematic for the 2009 field release
the fact that the dry ice was evaporating out of the boxes. Once the in-lab test was
done it was time to test the system out in the field.
First 2009 Field Experiment
A month long field experiment was conducted at the ZERT field site beginning on
July 12th and running until August 19th. An injection rate of 0.2 tons per day was
used during the two CO2 releases during the summer of 2009. The injection period of
CO2 occurred from July 14th to August 12th. The system was positioned as follows:
box 1, which houses the 1 meter free space absorption cell, was placed directly over
the injection pipe. Box two was also placed directly over the pipe but was positioned
1 meter west of box 1. Box 3 was positioned 1 meter away from the injection well and
box 4 was positioned 2 meters away from the well. During the month long injection
box 2 and box 3 malfunctioned and were not able to collect any useable data. The
malfunction was due to the fact that insufficient power was delivered to the detectors,
this will be futher discussed later in this section. Data from box 1 and box 4 were
collected through out the experiment.
30
Figure 20: Results for an in-lab experimental test run using dry ice.
Figure 21 is the plot of the concentration levels measured by the 1 meter free space
cell located over the injections well (solid black line) and the 1 meter fiber cell located
2 meters away from the injection well (solid red line). The data collected shows that
the CO2 released from the pipe does not radiate very far from the injection well. The
absorption cell located over the pipe clearly shows an increase in CO2 about 5 days
after the beginning of CO2 injections. The CO2 concentration for the cell located over
the pipe stayed elevated throughout the injection period. The absorption cell located
2 meters away never measured elevated in CO2 concentrations. The only variation for
the fiber absorption cell is the daily diurnal cycle that occurs naturally due to plant
and microbial reparation. Though not fully functioning this experimental deployment
demonstrated that the migration of CO2 ejected from the underground well does not
radiate very far away from the pipe.
31
Figure 21: CO2 concentrations for the month long injection experiment. The solid
black line indicates measurements made over the injection well, while the solid red
line indicates measurements made 2 meters away from the pipe.
Second 2009 Field Experiment
Due to the unexpected malfunctions of two of the four sensors for the underground
system a second CO2 field experiment was conducted during the summer of 2009.
Before the experiment began again a few corrections were made to the system. One of
the main difficulties that occurred during the first 2009 experiment was that the signal
delivered to the underground detectors would decrease to the point that measurements
could not be made. The decrease in signal would occur days after the system would
be buried underground. After careful consideration it was concluded that water vapor
was condensing onto the end of the fiber causing interference in optical power being
delivered to the detector. The first alteration done to the system was that each of
the three fiber absorption cells were taken back into the lab and the ends of the fiber
were re-cleaved to achieve a clean and open end. To keep the ends free and open,
32
desiccant was placed inside each of the boxes to absorb any water vapor that can
clog the end of the fiber cells. The next alteration made to the system was to take 3
inch diameter PVC piping and directly connect the pipe to the outside of each of the
three fiber boxes. This insures that while burying the boxes the dirt will not cause
the BNC cables and fiber optical cable to become stressed at the connectors on the
boxes.
Once all of the changes to the system were made a second experiment was performed out at the ZERT field site. A one week injection of CO2 occurred from
September 8th to September 15th, 2009. The system was deployed with three of the
four sensors. The three sensors that were used was the 1 meter free space cell located
over the injection pipe, and two of the fiber cell one located over the pipe, and one
located 1 meter away from the pipe. All three of the sensors were located at a depth of
approximately 0.7 meters below ground. Figure 22 is a plot of the data collected over
the course of the experiment. The solid black line represents the CO2 concentrations
made using the 1 meter long free space cell located over the injection pipe. The solid
red line represents the 1 meter fiber cell that was located over the pipe. The solid
green line represents the 1 meter fiber cell located 21 meter away from the injection
well. The system was deployed September 5th in order to get background readings
of naturally occurring CO2 before the CO2 was released from the underground well.
The two vertical lines in Figure 22 represents the beginning of the CO2 injection and
the end of the injection of CO2 . An increase in CO2 levels was recorded about 24
hours after the injection started by the 1 meter free space cell, a few hours later the
1 meter fiber cell began detecting increasing levels of CO2 . Approximately 48 hours
after the injection of CO2 began the fiber sensor located 1 meter away from the injection well also saw in increase in CO2 . An unexplainable anomaly occurred during
the week long injection, all three sensors observed a decrease in CO2 concentration
33
Figure 22: CO2 concentrations for the week long injection experiment. The solid black
line indicates measurements made by the 1 meter free space cell over the injection
well while the solid red line indicates measurements made by the 1 meter fiber cell,
and the green line indicates the fiber cell 1 meter away from the pipe.
though the injection well recorded a constant flow rate. Though it is unknown where
the CO2 went having all three sensors agree shows that it was not an inaccuracy of
the underground system. After the CO2 was turned off it took approximately 6 days
for the CO2 levels in the soil to return to the background level CO2 concentration.
The second release experiment was extremely successful and demonstrated that an
array of fiber sensors could be deployed successfully.
34
CONCLUSION
An underground differential absorption monitoring system for monitoring CO2
concentrations has been built, tested, and has successfully been used to detected elevated CO2 concentrations from a test field site. The underground instrument uses a
tunable distributed feed back laser that is capable of tuning over several absorption
features around 2 µm to calculate the concentration level of CO2 . The field deployment in 2008 showed that the system could accurately detect CO2 that was leaking
from an underground source. It was shown that the underground differential absorption monitoring system could clearly distinguish between naturally occurring CO2
and an external source of CO2 . The 2008 field experiment also validated the use of
photonic bandgap fiber as an absorption cell. The field experiment for 2009 expanded
on the results that were produced the year earlier and an array of fiber sensors were
deployed along with one of the free space absorption cells previously used. The data
collected in 2009 further demonstrated that the system could be used to continuously
measure CO2 concentration.
Future Work
The underground differential absorption monitoring system was capable of detecting CO2 leaking from an underground fault or fracture site. Each year the system
has improved to become an array of smaller sensors. In the future, in order to make
the system become more deployable at a real world sequestration sites some improvements are needed including expanding the underground system from 4 sensors to an
array of one hundred, to cover a much more extensive area, an optical switch with
many more optical outputs is needed. Once the ability to deliver light to many more
35
underground systems is achieved then the next concept that should be explored is to
make the underground portion of the system even smaller so that virtually no ground
is disturbed. By decreasing the size of the underground boxes the concept of having
one hundred sensors is more manageable and could be easily positioned.
In order to make the system more commercially friendly considerations need to be
made to make the system more automated. First this would entail making the above
ground portion of the system completely temperature insensitive. The laser portion
of the system is subject to power and noise issues when the temperature out in the
field becomes hot. Previously this was fixed by propping the boxes open that house
the laser and computer system so that heat inside the boxes could escape. Because
weather is variable in Montana, a constant awareness of the weather was needed so
that the boxes could be closed if any rain was impending. Because power is limited in
the field only a small thermal electric cooler was tried in the summer of 2008 but the
cooling output of the system did not relate well to the size of the box and temperature
still affected the laser. By using solar cells a larger cooling system could be considered
without having to consider the power used out in the field.
Another way to automate the system would be to have remote access to the
computer system out in the field. Only recently has internet access been available
out in the field and it would be beneficial to be able to monitor the system at all
time for power fluctuations and temperature changes. In the end the future of the
underground system is to expand the point source measurements to a large grid that
can be automated and run with minimal human interaction.
36
REFERENCES
[1] “Monthly average carbon dioxide concentration,” May 2007. [Online]. Available:
http : //scrippsco2.ucsd.edu/graphicsg allery/maunal oar ecord/mlor ecord.html
[2] L. S. Rothman, A. Barbe, D. C. Benner, L. R. Brown, C. Camy-Peyret,
M. Carleer, K. Chance, C. Clerbaux, V. Dana, V. M. Devi, A. Fayt, J. Flaud,
R. Gamache, A. Goldman, D. Jacquemart, K. Jucks, W. J. Lafferty, J. Mandin,
S. Massie, V. Nemtchinov, D. Newnham, A. Perrin, C. Rinsland, J. Schroeder,
K. Smith, M. Smith, K. Tang, R. A. Toth, J. V. Auwera, P. Varanasi, and
K. Yoshino, “The hitran molecular spectroscopic database: edition of 2000 including updates through 2001,” Journal of Quantitative Spectroscopy and Radiative Transfer, vol. 82, pp. 5–44, March 2003.
[3] J. Lewicki, “Rain data collected by instruments operated by dr. jennifer lewicki
of berkeley national labs,” 2008.
[4] C. M. Oldenburg, J. L. Lewicki, L. Dobeck, and L. Spangler, “Modeling gas
transport in the shallow subsurface during the zert co2 release test,” Transport
in Porous Media, 2009.
[5] Z. Li, M. Dong, S. Li, and S. Huang, “Co2 sequestration in depleted oil and
gas reservoirs-caprock characterization and storage capacity,” Energy Conversion
and Management, vol. 47, pp. 1372–1382, 2006.
[6] K. Masarie and P. T. Tans, “Extension and integration of atmosphere carbon
dioxide data into a globally consistent measurement record,” Journal of Geophysical Research, vol. 100, pp. 11 593–11 610, June 1995.
[7] P. Tans, “How can global warming be traced to co2?” Scientific American, vol.
295, no. 6, p. 124, Dec 2006.
[8] M. Scheffer, V. Brovkin, and P. M. Cox, “Positive feedback between global warming and atmospheric co2 concentration inferred from past climate change,” Geophysical Research Letters, vol. 33, p. L10702, 2006.
[9] J. Hansen, “Defusing the global warming time bomb,” Scientific American, vol.
290, no. 3, p. 68, 2004.
[10] T. Xu, “Co2 geological sequestration,” Lawrence Berkeley National Laboratory,
vol. Paper LBNL-56644 JArt, November 2004.
37
[11] S. Whittaker, K. Kreis, T. Davis, Z. Hajnal, T. Heck, L. Penner, H. Qing, and
B. Rostron, “Characterizing the geologic container at the weyburn field for subsurface co2 storage associated with enhanced oil recovery,” Proceedings of the
Diamond Jubilee convention of the Canadian Society of Petroleum Geologists,
2002.
[12] H. J. Herzog, “What future for carbon capture and sequestration?” American
Chemical Society, vol. 35, no. 7, pp. 148 A–153 A, April 2001, massachusetts
Institute of Technology.
[13] S. Whittaker, “Geological storage of greenhouse gases: The iea weyburn co2
monitoring and storage project,” Canadian Society of Petroleum and Geologists
Reservoir, vol. 31, no. 8, Sep 2004.
[14] S. Hovorka, S. Benson, C. Doughty, B. Freifeld, S. Sakurai, T. Daley, Y. Kharaka,
M. Holtz, R. Trautz, H. Nance, L. Myer, and K. Knauss, “Measuring permanence
of co2 storage in saline formations: the frio experiment,” Environmental Geosciences, vol. 13, no. 2, June 2006.
[15] T. Torp and J. Gale, “Demonstration storage of co2 in geological reservoirs: The
sleipner and sacs projects,” Energy, vol. 29, 2004.
[16] A. Mathieson, I. Wright, D. Roberts, and P. Ringrose, “Satellite imaging to
monitor co2 movement at krechba, algeria,” Science Direct, vol. 1, 2009.
[17] S. M. Benson, E. Gasperikova, and G. M. Hoversten, “Monitoring protocols and
life-cycle costs for geologic storage of carbon dioxide,” Proceedings of the 7th
International conference on greenhouse Gas control Technologies, 2005.
[18] S. Humphries, A. Nehrir, C. Keith, K. Repasky, L. Dobeck, J. Carlsten, and
L. Spangler, “Testing carbon sequestration site monitor instruments using a controlled carbon dioxide release facility,” Applied Optics, vol. 47, no. 4, February
2008.
[19] K. S. Repasky, S. Humphries, and J. L. Carlsten, “Differential absorption measurements of carbon dioxide using a temperature tunable distributed feedback
diode laser,” Review of Scientific Instruments, vol. 77, 2006.
[20] nanoplus. 2 micron dfb laser. [Online]. Available: http://www.nanoplus.com
[21] millipore corperation.
www.millipore.com
catalog
number
falp14250.
[Online].
Available:
[22] Y. Li, C. Wang, M. Hu, B. Liu, X. Sun, and L. Chai, “Photonic bandgap fibers
based on a composite honeycomp lattice,” IEEE Photonics Technology Letters,
vol. 18, no. 1, January 2006.
38
[23] R. Thapa, K. Knabe, K. Corwin, and B. Washburn, “Arc fusion splicing of
hollow-core photonic bandgap fibers for gas-filled fiber cells,” Optics Express,
vol. 14, no. 21.
[24] Y. Hoo, W. Jin, C. Shi, H. Ho, D. Wang, and S. Ruan, “Design and modeling of
a photonic crystal fiber gas sensor,” Applied Optics, vol. 42, no. 18, june 2003.